The Big Cloud Trajectory

While AWS and Azure still hold the top spots on the big cloud market share leaderboards, I believe Google Cloud is stacking up the integrated features that will set it truly apart over the next 5 years.

Gemini 3 is a winner.

The TPU is real differentiation to the circular NVIDIA economy.

Warren Buffett agrees.

And Firebase, though only a small part of Google Cloud, has continued to receive strong investment from Google, offering best-in-class, extensible, and unique tools for building innovative apps quickly and robustly.

In this lengthy post, we'll make the argument for Firebase and GCP in 2026. In future posts, we'll dive deep into many of the features and tools in GCP that can help you build truly innovative MVPs and add next-level features to existing apps.

Why Firebase?

In 2025, you can't really hold anyone's attention without AI in your product pitch. The simple truth though is very few, if any, of us are going to build an LLM to compete with the likes of OpenAI, Anthropic, and Google. If we want to innovate, we'll be embedding, integrating, and RAG'ing our way into differentiated AI features in our apps.

So our focus should remain, as always, on the real prize: applying our craft to build solutions that help people.

I'm a strong believer that reliability is feature #1. The principle, popularized by Google's SRE philosophy, states that users won't use solutions they can't trust.

Yet as software engineers, we know how easy it is to ship unreliable shit. Accidental complexity comes for us all, often late at night or attached to anxious or angry voices. And no amount of LLM-guidance can save us from our solution's inherent complexity.

So we have to manage it. Type safety, testing, and good design patterns get us a long way. But if our solutions exist in the cloud, we have to integrate with a host of platforms, databases, message queues, event stores, object stores, auth systems, LLM's, & networks. We have to scale up, scale down, migrate, rollback, & observe.

Many of us have the talent to craft our own solutions, with ample glue code, to any and all of these complex challenges. And if our app one day earns its late-stage capitalism unicorn cutie mark, we'll have a whole team of ponies to help us pull it off.

But if we're trying to build an innovative app from scratch with a small team, we need to be laser-focused on building the reliable features that provide a new and novel solution to our users' problems.

For that, I believe Firebase is the absolute best option you can choose in 2026.

App Hosting

We'll start our persuasive tour of Firebase with the recently GA'd release of Firebase App Hosting.

Even if you're shipping a desktop or mobile app, your product needs a web presence that supports SEO. And in 2026, you really can't do better than NextJS for building it.

The LLMs have all enjoyed olympic levels of training on React & NextJS, and I consistently get Gemini 3 & Claude to generate & update large parts of an implementation flawlessly. It's also very straightforward to deploy on App Hosting.

If you want the standard enterprise option, Angular is supported as well, with many more frameworks supported by the community.

App Hosting enhances the already excellent conveniences of Firebase Hosting, making it dead simple to put a global scale https-terminating loadbalancer and caching CDN in front of your web app.

App Hosting's Cloud Build integration provides a built-in CI/CD solution, which saves you a lot of trouble setting up your pipeline early on.

In future posts, we'll dive deeper into how you can slowly extend with Cloud Build & Cloud Deploy as you integrate more distributed services into your app.

Easy Authentication with Declarative Authorization that Integrates Across the Stack

A giant component of reliability is security. Your users sure as hell can't trust software that airs their laundry to the world.

Firebase Auth provides a robust, integrated solution on both ends of application security: authentication & authorization.

Whether you want to use only traditional email/password, email link, SMS, or OAuth 2.0 login flows, or all of the above, authentication is orders of magnitude less annoying to implement with Firebase than rolling your own. You will get it wrong if you attempt it, so I recommend you wait until you have your team of ponies before biting that bullet.

Authorization is where most of the application security fails occur. Given your data model is almost always multidimensional, and authorization can only safely be decided on the server-side, the gating authorization logic you have to write in front of every query is almost guaranteed to introduce some security holes.

Firebase, uniquely, provides statically analyzable authorization layers that fully integrate into every storage product. Instead of imperative auth logic that must be sprinkled all over your services, with Firebase you define schema and authorization together at once. This is incredibly powerful, and while still complex to get right, it wittles down the problem to just the inherent complexity of your data model, not all the accidental complexity that emerges from your ever-evolving API layers.

In future posts we'll be diving deeply into the many details of Firebase Auth and explore how it integrates with other Firebase layers.

Realtime Documents, Easy Presence, Secure GraphQL Postgres, & Simple Object Storage

I was a huge fan of Firebase even before it got acquired by Google. I think there's still a lot of value to be squeezed from realtime-by-default application concepts, and Firebase arguably provides the most complete set of tools to build them at scale.

Firebase's Firestore gives you realtime documents, transactional updates, and enough querying capability to cover most realtime app use cases.

Firebase's Realtime Database is what Google bought back in 2014. Firebase strongly recommends you use it sparingly, but it remains the only option for handling user presence in the Firebase toolkit. I won't argue this is at all elegant, but the capabilities you need are all there.

Firebase's Data Connect is bright and shining new, going GA back in April 2025. It provides a Firebase Auth integrated GraphQL endpoint for a Cloud SQL Postgres database. It's an excellent option for when your app needs full relational database capabilities, and can compliment Firestore quite effectively.

Finally, Firebase's Object Storage gives you a simple, secure, and cost-effective way to store and serve both user generated and AI generated files. It comes with its own Auth integrated declarative authorization layer, making it even more effective for application development than using Cloud Storage directly.

As we'll see in future posts, all these data layers have their place, and we'll explore the details of how best to use them, both separately and in combination.

Agentic AI with Genkit supported by Vectors in Firestore & Postgres

And finally to the giant elephant in everyone's virtual dev room: What are our AI features & how do we build them?

Firebase's answer to both questions is AI Logic and Genkit.

AI Logic is a client sdk layer that allows you to go straight to Gemini without needing server-side logic.

Going 1.0 in January 2025, Genkit is a library for inserting any set of LLMs into about any workflow your app may need. It's fully open source, continuing Google's practice of putting GCP features on open source foundations. If you're a Typescript developer, Genkit will finally give you confidence that your deterministic code can robustly wrap AI's indeterminism.

And both Firestore and Data Connect provide vector embedding support that integrates with Genkit, making it easy to build up your own RAG features that stay in sync with your data model.

We'll be spending lots of time with Genkit in future posts, so stay tuned!

Antigravity & Firebase Studio

And, finally, we can't evaluate the platform without also evaluating the platform-focused dev environments. Both Antigravity and Firebase Studio provide a VSCode based dev environment with differing levels of Gemini integration. Pairing either with the Firebase MCP server gives you quite a lot of functionality, and we'll dive into how best to maximize their uses in future posts.

The Case is Compelling

We haven't even touched on many of Firebase's features, yet the argument for Firebase is a winning one. We will be shifting to vlog format for the rest of this series, which will allow us to do full walkthroughs of the Firebase and GCP consoles as we edit and deploy tutorial implementations.

If you need to move your project forward, consider Firebase. And if you need a guide to ensure your success, work with us.